Original scientific paperThis paper deals with analytical and simulation approach for choice of activation functions and instead of a number of nodes for a class of neural network controllers for frequency control of thermal power systems. Neural network update laws are derived via Lyapunov like stability analysis. When number of nodes is fixed, then simulation analysis is conducted to find the best performer activation function. Best performance is chosen using integral error criteria and proper statistical tests.
Key words:Choice of activation functions, Number of nodes, Neural network control Dizajn adaptivnog regulatora s neuronskim mrežama za regulaciju frekvencije u elektroenergetskom sustavu s toplinskim turbinama.Članak opisuje analitičko-simulacijski pristup izboru aktivacijskih funkcija i brojačvorova za klasu regulatora s neuronskim mrežama za regulaciju frekvencije u elektroenergetskom sustavu s toplinskim turbinama. Zakoni učenja neuronske mreže su izvedeni kroz analizu stabilnosti primjenom funkcije Ljapunova. Nakon što je odreîen brojčvorova, proveden je simulacijski postupak za odreîivanje najbolje aktivacijske funkcije. Najbolja funkcija za dani regulator je odreîena korištenjem kriterija integralne pogreške i statističkim testiranjem.Ključne riječi: izbor aktivacijskih funkcija, brojčvorova, regulacija neuronskim mrežama
The paper deals with analysis of turbine governor positioning system of hydroelectric power plant regarding nonlinear resonance jnmp. The term "resonance jump" is used in the case of a sudden jump of amplitude and/or frequency and/or phase of a periodic output signal of a nonlinear system. Resonance jnmp cannot occur if excitation is such that the response of the system is transient and cannot be defined by solving nonlinear differential equations. Resonance jumps can cause severe damage and the mechanical, hydraulic and electrical systems with the saturation should be analyzed with the respect to the possibility of occurrence of resonance jump. In the paper the analysis of the water turbine governor positioning system with the respect to the occurrence of resonance jump is described. Two methods of analyzing system with the respect to resonance jumps are described: the simulation method and the analytical method. When using the simulation method, the analysis of the variations of important parameters, including dead zone is performed and the simulation procedures are detailed. Then, the confirmation of the results of simulation is given by the analytical analysis method based on the describing function approach. At the end, the fuzzy regulator for avoidance of nonlinear resonance jump in turbine governor positioning system is involved.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.